public FormattingAssemblerSettings()
{
RemoveStyleNamesFromParagraphAndRunProperties = true;
ClearStyles = true;
OrderElementsPerStandard = true;
CreateHtmlConverterAnnotationAttributes = true;
RestrictToSupportedNumberingFormats = false;
RestrictToSupportedLanguages = false;
ListItemRetrieverSettings = new ListItemRetrieverSettings();
}
private static void InitializeListItemRetrieverForPart(WordprocessingDocument wordDoc, OpenXmlPart part, ListItemRetrieverSettings settings)
{
var mainXDoc = part.GetXDocument();
var numPart = wordDoc.MainDocumentPart.NumberingDefinitionsPart;
if (numPart == null)
{
return;
}
var numXDoc = numPart.GetXDocument();
var stylesPart = wordDoc.MainDocumentPart.StyleDefinitionsPart;
if (stylesPart == null)
{
return;
}
var stylesXDoc = stylesPart.GetXDocument();
var rootNode = mainXDoc.Root;
InitializeListItemRetrieverForStory(numXDoc, stylesXDoc, rootNode);
var textBoxes = mainXDoc
.Root
.Descendants(W.txbxContent);
foreach (var textBox in textBoxes)
{
InitializeListItemRetrieverForStory(numXDoc, stylesXDoc, textBox);
}
}
private static void InitializeListItemRetriever(WordprocessingDocument wordDoc, ListItemRetrieverSettings settings)
{
foreach (var part in wordDoc.ContentParts())
{
InitializeListItemRetrieverForPart(wordDoc, part, settings);
}
#if false
foreach (var part in wordDoc.ContentParts())
{
var xDoc = part.GetXDocument();
var paras = xDoc
.Descendants(W.p)
.Where(p =>
p.Annotation <ListItemInfo>() == null);
if (paras.Any())
{
Console.WriteLine("Error");
}
}
#endif
}
public static string RetrieveListItem(WordprocessingDocument wordDoc, XElement paragraph, ListItemRetrieverSettings settings)
{
if (wordDoc.MainDocumentPart.NumberingDefinitionsPart == null)
{
return(null);
}
var listItemInfo = paragraph.Annotation <ListItemInfo>();
if (listItemInfo == null)
{
InitializeListItemRetriever(wordDoc, settings);
}
listItemInfo = paragraph.Annotation <ListItemInfo>();
if (!listItemInfo.IsListItem)
{
return(null);
}
var numberingDefinitionsPart = wordDoc
.MainDocumentPart
.NumberingDefinitionsPart;
if (numberingDefinitionsPart == null)
{
return(null);
}
StyleDefinitionsPart styleDefinitionsPart = wordDoc
.MainDocumentPart
.StyleDefinitionsPart;
if (styleDefinitionsPart == null)
{
return(null);
}
var numXDoc = numberingDefinitionsPart.GetXDocument();
var stylesXDoc = styleDefinitionsPart.GetXDocument();
var paragraphLevel = GetParagraphLevel(paragraph);
var lvl = listItemInfo.Lvl(paragraphLevel);
string lvlText = (string)lvl.Elements(W.lvlText).Attributes(W.val).FirstOrDefault();
if (lvlText == null)
{
return(null);
}
var levelNumbersAnnotation = paragraph.Annotation <LevelNumbers>();
if (levelNumbersAnnotation == null)
{
throw new OpenXmlPowerToolsException("Internal error");
}
int[] levelNumbers = levelNumbersAnnotation.LevelNumbersArray;
string languageIdentifier = GetLanguageIdentifier(paragraph, stylesXDoc);
string listItem = FormatListItem(listItemInfo, levelNumbers, GetParagraphLevel(paragraph),
lvlText, stylesXDoc, languageIdentifier, settings);
return(listItem);
}
private static string FormatListItem(ListItemInfo lii, int[] levelNumbers, int ilvl,
string lvlText, XDocument styles, string languageCultureName, ListItemRetrieverSettings settings)
{
string[] formatTokens = GetFormatTokens(lvlText).ToArray();
XElement lvl = lii.Lvl(ilvl);
bool isLgl = lvl.Elements(W.isLgl).Any();
string listItem = formatTokens.Select((t, l) =>
{
if (t.Substring(0, 1) != "%")
{
return(t);
}
int indentationLevel;
if (!Int32.TryParse(t.Substring(1), out indentationLevel))
{
return(t);
}
indentationLevel -= 1;
if (indentationLevel >= levelNumbers.Length)
{
indentationLevel = levelNumbers.Length - 1;
}
int levelNumber = levelNumbers[indentationLevel];
string levelText = null;
XElement rlvl = lii.Lvl(indentationLevel);
string numFmtForLevel = (string)rlvl.Elements(W.numFmt).Attributes(W.val).FirstOrDefault();
if (numFmtForLevel == null)
{
var numFmtElement = rlvl.Elements(MC.AlternateContent).Elements(MC.Choice).Elements(W.numFmt).FirstOrDefault();
if (numFmtElement != null && (string)numFmtElement.Attribute(W.val) == "custom")
{
numFmtForLevel = (string)numFmtElement.Attribute(W.format);
}
}
if (numFmtForLevel != "none")
{
if (isLgl && numFmtForLevel != "decimalZero")
{
numFmtForLevel = "decimal";
}
}
if (languageCultureName != null && settings != null)
{
if (settings.ListItemTextImplementations.ContainsKey(languageCultureName))
{
var impl = settings.ListItemTextImplementations[languageCultureName];
levelText = impl(languageCultureName, levelNumber, numFmtForLevel);
}
}
if (levelText == null)
{
levelText = ListItemTextGetter_Default.GetListItemText(languageCultureName, levelNumber, numFmtForLevel);
}
return(levelText);
}).StringConcatenate();
return(listItem);
}
private static void AssembleListItemInformation(WordprocessingDocument wordDoc, ListItemRetrieverSettings settings)
{
foreach (var part in wordDoc.ContentParts())
{
XDocument xDoc = part.GetXDocument();
foreach (var para in xDoc.Descendants(W.p))
{
ListItemRetriever.RetrieveListItem(wordDoc, para, settings);
}
}
}
public static void AssembleFormatting(WordprocessingDocument wDoc, FormattingAssemblerSettings settings)
{
FormattingAssemblerInfo fai = new FormattingAssemblerInfo();
XDocument sXDoc = wDoc.MainDocumentPart.StyleDefinitionsPart.GetXDocument();
XElement defaultParagraphStyle = sXDoc
.Root
.Elements(W.style)
.FirstOrDefault(st => st.Attribute(W._default).ToBoolean() == true &&
(string)st.Attribute(W.type) == "paragraph");
if (defaultParagraphStyle != null)
fai.DefaultParagraphStyleName = (string)defaultParagraphStyle.Attribute(W.styleId);
XElement defaultCharacterStyle = sXDoc
.Root
.Elements(W.style)
.FirstOrDefault(st => st.Attribute(W._default).ToBoolean() == true &&
(string)st.Attribute(W.type) == "character");
if (defaultCharacterStyle != null)
fai.DefaultCharacterStyleName = (string)defaultCharacterStyle.Attribute(W.styleId);
XElement defaultTableStyle = sXDoc
.Root
.Elements(W.style)
.FirstOrDefault(st => st.Attribute(W._default).ToBoolean() == true &&
(string)st.Attribute(W.type) == "table");
if (defaultTableStyle != null)
fai.DefaultTableStyleName = (string)defaultTableStyle.Attribute(W.styleId);
ListItemRetrieverSettings listItemRetrieverSettings = new ListItemRetrieverSettings();
AssembleListItemInformation(wDoc, settings.ListItemRetrieverSettings);
foreach (var part in wDoc.ContentParts())
{
var pxd = part.GetXDocument();
FixNonconformantHexValues(pxd.Root);
AnnotateWithGlobalDefaults(wDoc, pxd.Root, settings);
AnnotateTablesWithTableStyles(wDoc, pxd.Root);
AnnotateParagraphs(fai, wDoc, pxd.Root, settings);
AnnotateRuns(fai, wDoc, pxd.Root, settings);
}
NormalizeListItems(fai, wDoc, settings);
if (settings.ClearStyles)
ClearStyles(wDoc);
foreach (var part in wDoc.ContentParts())
{
var pxd = part.GetXDocument();
pxd.Root.Descendants().Attributes().Where(a => a.IsNamespaceDeclaration).Remove();
FormattingAssembler.NormalizePropsForPart(pxd, settings);
var newRoot = (XElement)CleanupTransform(pxd.Root);
pxd.Root.ReplaceWith(newRoot);
part.PutXDocument();
}
}
private static void InitializeListItemRetrieverForPart(WordprocessingDocument wordDoc, OpenXmlPart part, ListItemRetrieverSettings settings)
{
var mainXDoc = part.GetXDocument();
var numPart = wordDoc.MainDocumentPart.NumberingDefinitionsPart;
if (numPart == null)
return;
var numXDoc = numPart.GetXDocument();
var stylesPart = wordDoc.MainDocumentPart.StyleDefinitionsPart;
if (stylesPart == null)
return;
var stylesXDoc = stylesPart.GetXDocument();
var rootNode = mainXDoc.Root;
InitializeListItemRetrieverForStory(numXDoc, stylesXDoc, rootNode);
var textBoxes = mainXDoc
.Root
.Descendants(W.txbxContent);
foreach (var textBox in textBoxes)
InitializeListItemRetrieverForStory(numXDoc, stylesXDoc, textBox);
}
private static void InitializeListItemRetriever(WordprocessingDocument wordDoc, ListItemRetrieverSettings settings)
{
foreach (var part in wordDoc.ContentParts())
InitializeListItemRetrieverForPart(wordDoc, part, settings);
#if false
foreach (var part in wordDoc.ContentParts())
{
var xDoc = part.GetXDocument();
var paras = xDoc
.Descendants(W.p)
.Where(p =>
p.Annotation<ListItemInfo>() == null);
if (paras.Any())
Console.WriteLine("Error");
}
#endif
}
public static string RetrieveListItem(WordprocessingDocument wordDoc, XElement paragraph, ListItemRetrieverSettings settings)
{
if (wordDoc.MainDocumentPart.NumberingDefinitionsPart == null)
return null;
var listItemInfo = paragraph.Annotation<ListItemInfo>();
if (listItemInfo == null)
InitializeListItemRetriever(wordDoc, settings);
listItemInfo = paragraph.Annotation<ListItemInfo>();
if (!listItemInfo.IsListItem)
return null;
var numberingDefinitionsPart = wordDoc
.MainDocumentPart
.NumberingDefinitionsPart;
if (numberingDefinitionsPart == null)
return null;
StyleDefinitionsPart styleDefinitionsPart = wordDoc
.MainDocumentPart
.StyleDefinitionsPart;
if (styleDefinitionsPart == null)
return null;
var numXDoc = numberingDefinitionsPart.GetXDocument();
var stylesXDoc = styleDefinitionsPart.GetXDocument();
var lvl = listItemInfo.Lvl(GetParagraphLevel(paragraph));
string lvlText = (string)lvl.Elements(W.lvlText).Attributes(W.val).FirstOrDefault();
if (lvlText == null)
return null;
var levelNumbersAnnotation = paragraph.Annotation<LevelNumbers>();
if (levelNumbersAnnotation == null)
throw new OpenXmlPowerToolsException("Internal error");
int[] levelNumbers = levelNumbersAnnotation.LevelNumbersArray;
string languageIdentifier = GetLanguageIdentifier(paragraph, stylesXDoc);
string listItem = FormatListItem(listItemInfo, levelNumbers, GetParagraphLevel(paragraph),
lvlText, stylesXDoc, languageIdentifier, settings);
return listItem;
}
private static string FormatListItem(ListItemInfo lii, int[] levelNumbers, int ilvl,
string lvlText, XDocument styles, string languageCultureName, ListItemRetrieverSettings settings)
{
string[] formatTokens = GetFormatTokens(lvlText).ToArray();
XElement lvl = lii.Lvl(ilvl);
bool isLgl = lvl.Elements(W.isLgl).Any();
string listItem = formatTokens.Select((t, l) =>
{
if (t.Substring(0, 1) != "%")
return t;
int indentationLevel;
if (!Int32.TryParse(t.Substring(1), out indentationLevel))
return t;
indentationLevel -= 1;
if (indentationLevel >= levelNumbers.Length)
indentationLevel = levelNumbers.Length - 1;
int levelNumber = levelNumbers[indentationLevel];
string levelText = null;
XElement rlvl = lii.Lvl(indentationLevel);
string numFmtForLevel = (string)rlvl.Elements(W.numFmt).Attributes(W.val).FirstOrDefault();
if (numFmtForLevel == null)
{
var numFmtElement = rlvl.Elements(MC.AlternateContent).Elements(MC.Choice).Elements(W.numFmt).FirstOrDefault();
if (numFmtElement != null && (string)numFmtElement.Attribute(W.val) == "custom")
numFmtForLevel = (string)numFmtElement.Attribute(W.format);
}
if (numFmtForLevel != "none")
{
if (isLgl && numFmtForLevel != "decimalZero")
numFmtForLevel = "decimal";
}
if (languageCultureName != null && settings != null)
{
if (settings.ListItemTextImplementations.ContainsKey(languageCultureName))
{
var impl = settings.ListItemTextImplementations[languageCultureName];
levelText = impl(languageCultureName, levelNumber, numFmtForLevel);
}
}
if (levelText == null)
levelText = ListItemTextGetter_Default.GetListItemText(languageCultureName, levelNumber, numFmtForLevel);
return levelText;
}).StringConcatenate();
return listItem;
}
